Self-supporting Angles in Large Scale Additive Manufacturing

Self-supporting Angles in Large Scale Additive Manufacturing

Trivial and Non-Trivial Findings

By Austin Schmidt, Additive Engineering Solutions

At AES, we are always trying to improve our production methods in order to better serve our customers. In order to create the best product possible, our team must continually innovate and iterate on production methods. At AES, we are able to research and perform these studies to help foster productive growth in the large scale A.M. industry.

Spend any time operating or designing projects for a 3D printer* and you will quickly learn about the function & limitations of self-supporting angles, as well as the importance of properly designed angles throughout the geometry of your project. Self-supporting angles are the angles at which a 3D printer can build overhangs on successive layers without the need to provide support material below the layer being printed. Building a self-supporting structure with A.M. is like building an upside down staircase, as shown in the figure below.

While it is possible for each layer to overhang a bit from the layer below, step out too far and gravity will pull the plastic down. This will cause the build to fall apart and fail. For the sake of this discussion, we will consider the angles in question to be measured from vertical.


With Cincinnati’s BAAM®, along with other forms of large format 3D printers such as Thermwood’s LSAM®, it is important to design prints with self-supporting features in mind as none of these machines offer a support option that is easily removed from the print. Abiding by the self-supporting angle will result in the highest likelihood of print success.

For traditional Fused Filament Fabrication (FFF) processes such as Stratasys FDM® along with desktop printing, self-supporting angles have been well studied. With a little bit of online research, guidelines can easily be found. However, large scale 3D printing is a relatively new process and such research in the area is minimal. Additionally, self-supporting angles can vary by material. Since large scale A.M. uses injection molding pellets as the feedstock, the variety of material is nearly limitless. However, each material will have its own set of characteristics which may affect the maximum self-supporting angle.

Recently at AES, we needed to determine the self-supporting angle of a particular polycarbonate in order to effectively produce a part for a customer. At the onset of the project, we thought an angle of 45 degrees would be acceptable in this particular situation. However, after some initial printing, we realized that this was too aggressive of an angle. At that point, we decided to conduct our own study to determine the maximum self-supporting angle.


Here is a look at the geometry we used in the study. The overall height of the part varied a bit based on the self-supporting angle. However, the total flange width at top was held constant at 1.5” and the initial wall thickness was .5”. Additionally, at least 1.5” was printed before the flange started and at least 2 layers, or .2” were built vertically on top of the flange. All dimensions shown below are in inches.


Machine testing Parameters:

Machine: Cincinnati BAAM®

Material: Polycarbonate Reinforced with 20% Glass Fiber (SABIC’s THERMOCOMP™ compound DF004)

Nozzle Diameter: .20”

Bead Width Setting: .23”

Layer height Setting: .10”

Pump Speed: 277 rpm

Travel Speed: 18.228 in/s

Inset Value: 3

Skins: None

Infill: None

Initial Start/Stop Strategy: Random

Results: Test #1

The failing of several pieces at a 45 degree angle were the driving force to initiate these tests. For this first test, we printed 4 parts simultaneously, with each having a different self-supporting angle. For this test, we chose 25, 30, 35 and 40 degrees. Based on our previous experience with the 45 degree angles and some analysis of the print slices, we felt fairly confident that the 25 and 30 degree angles would print successfully while the 35 and 40 degree angles may be a bit more questionable. The results of this test are shown below.


Much to our surprise, the 25 degree and 40 degree angles built correctly, while the 30 and 35 degree angles failed. This was a very confounding result, as we would have expected a transition from failure to success as the angle decreased.

Results: Test #2

After reviewing the results of Test #1, we decided to repeat the test with one key parameter change. In our first test, as well as with the failed parts that led us to this trial, we had the “Start/Stop” set to “Random”. This means that in our test geometry, the starting and stopping location of each bead printed was at a different location on the circumference of the part and different on each layer. Initially, we did not think this was going to be a very important parameter in this study. However, after closely watching the printing occur in the first test, we made two observations

  1. Random Smart/Stop causes the inconsistency with the layer time across the parts. With this setting enabled, the path planning for the extruder tends to jump around between pieces and does not result in a consistent layer time between beads. For example, when printing layers we often saw it print piece 1, piece 2, piece 3, piece 4 which would be normal, but once the table lowered for the next layer, the extruder would work backwards starting with Piece 4 then Piece 3, Piece 2 and Piece 1. This meant the layer to layer tme for Piece 4 was significantly different from Piece 1.
  2. Second, we noticed that when one of the parts started to fail, it was due to the interior most bead failing to adhere to the supporting layer below the start point of the build path. Once this bead fails to adhere, it is dragged off the part and it becomes impossible to recover the build. The printing of this piece would ultimately fail.

For Test #2, we decided to change this parameter to what is commonly referred to as “Optimized Start/Stop”. This start/stop strategy gives the user some control over where the start/stop location is oriented on the part but, more importantly in our case, it makes the extruder start and stop at the same X-Y location on every layer.

This does two things. First, it makes the path planning more consistent. On any given layer, the machine always printed the pieces in the same order. This made for more consistent layer time. Second, at the “Start/Stop” location, there is a tendency for the material to “pool” a bit. The result is a small section that is usually a bit wider than the rest of the bead. Technically, this would be considered a build defect and effort is given to try to minimize this size of this Start/Stop defect. However, in this case, we believe that the extra surface area of the start/stop pad on the layer below may have given the bead above a better place to anchor at its start. With the increased surface area at the start of the bead, the bead was more likely to adhere and continue as a successful print.

With this parameter changed, we repeated the Test #1 file holding all parameters other than “Start/Stop” the same. As the picture shows below, all four pieces built correctly.


Results: Test #3

Based on our new-found success in Test #2, we decided to see how big of an angle we could print using the optimized Start/Stop method. For this test, we decided to print 45, 50, and 55 degree angle pieces. Since we wanted to make sure the layer times were similar across all tests, and because we were interested in the original part specification of 45 degrees, we decided to print two at that angle. As shown in the photo below, all 4 of these pieces failed.


Opportunities for Additional Study

This study was conducted to solve an immediate problem facing AES and the results gathered above were enough to give us the confidence we needed to produce the required parts. Given additional time and resources, our team sees opportunities for additional research in this space.

First and foremost, it would be wise to repeat this study with a statistically significant number of test pieces. Cost and time were factors in this study that prevented us from running additional samples.

There are two parameters in this study that, if varied, could also lead to different results. The first variable is melt temperature in relation to layer time. Increasing or decreasing the melt temperature varies the amount of heat put into the plastic. This, in turn, affects the minimum and maximum layer time of the print. If too much heat is put into the bead relative to the layer time (amount of time the material has to cool) subsequent layers may have a tendency to slump causing a failed print. This phenomenon can be exacerbated when building self-supporting angles. Similarly, if the layer cools too much before the subsequent layer is applied, layer to layer adhesion may be poor causing the build to fail. The second parameter is material selection. This study covers SABIC’s THERMOCOMP™ compound DF004, a 20% glass fiber reinforced polycarbonate. While other polycarbonates may react similarly, other materials (Nylon resin, ABS resin, ULTEM™ resin, VALOX™ resin, NORYL™ resin, PPS resin, PPSU resin, etc.) may result in different range of manufacturable angles.

Finally, Alex Roschli from Oak Ridge National Laboratory made an interesting observation regarding the geometry of the part. The flange in question is an inwardly built flange. Results may differ if the flange was built outward. On the inward built flange in this study, there is constant tension pulling on the bead as it is being deposited. On an inward built flange, a component of the tension force is pulling the bead toward the center of the circle, or off of the layer below. On an outwardly built flange, the same component of tension would be pulling bead back onto the already printed surface.


From these tests we draw two conclusions when printing this particular material with the parameters given above.

  • First, we would recommend the Optimized Start/Stop when building any kind of self-supporting feature.
  • Second, we would not recommend using any angle larger than 40 degrees until we have additional results that would prove steeper angles are acceptable.

As we mentioned above, the sample size of parts run for this test are small, so we would caution anyone looking to make definitive answers based off this study. However, it did provide us with results that we will be able to take forward on future projects and we hope it provides you with additional insight as you take new A.M. projects.

A special thanks goes out to Clark Patterson, Andrew Bader, John Sprouse, and Alex Roschli, all of whom contributed to this study.

SABIC and brands marked with ™ are trademarks of SABIC.

*Most SLS 3D printing does not require supports in the printing process.

Leave a Reply

Your email address will not be published. Required fields are marked *

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>